IRC logs for #openttd on OFTC at 2025-09-18
⏴ go to previous day
00:01:25 *** firepup650 has joined #openttd
02:31:22 *** WormnestAndroid has quit IRC (Read error: Connection reset by peer)
02:31:25 *** WormnestAndroid has joined #openttd
02:31:32 *** WormnestAndroid has quit IRC (Read error: Connection reset by peer)
02:31:34 *** WormnestAndroid has joined #openttd
02:31:38 *** WormnestAndroid has quit IRC (Read error: Connection reset by peer)
02:31:38 *** WormnestAndroid has joined #openttd
03:00:05 *** herms has quit IRC (Quit: bye)
03:49:10 *** Zathras has joined #openttd
03:52:36 *** Zathras_11 has quit IRC (Ping timeout: 480 seconds)
04:22:22 *** dh1 has quit IRC (Ping timeout: 480 seconds)
04:37:52 <DorpsGek> - Update: Translations from eints (by translators)
05:50:43 *** dh1 has quit IRC (Ping timeout: 480 seconds)
07:25:07 *** dh1 has quit IRC (Ping timeout: 480 seconds)
07:46:16 <peter1138> The Shai-Hulud npm worm has now reached 500+ packages...
07:46:56 <LordAro> i'd have thought npm would've locked it all down by now
07:47:25 <LordAro> they're normally fairly responsive about these things, even if they never do anything about the underlying issue
07:49:59 *** gelignite has joined #openttd
07:50:29 *** reldred has joined #openttd
07:52:54 <mnhebi> Blessed be the maker...
08:16:19 <peter1138> Hmm, survey says plenty of players use non-integer interface scale.
08:16:54 <reldred> I'd definitely love if I could use 3x
08:17:48 <peter1138> Which is non-integer when using a baseset with 2x or 4x sprites.
08:18:37 <reldred> Yee, it's not ideal with 1x either
08:18:45 <reldred> 2x with lots of padding
08:20:23 <reldred> For my 34" 3440x1440 panel 2x is fine, but I've got a 2560x1440 panel on my gpd winmax2 and that's 9-10". 4x is too cramped, 2x is eye squinting hunchback teritory
08:22:54 <peter1138> I did try linear-interpolation scaing but predictably it looks like shit.
08:23:17 <peter1138> But then again, people play Doom like that some how.
08:24:05 *** jfkuayue has joined #openttd
08:24:05 <jfkuayue> thanks all for help!
08:25:02 <peter1138> Although I use dsda-doom, which is apparently opinionated enough to not have that possibility, so it's all nice & crispy.
08:25:15 *** tabytac has joined #openttd
08:25:15 <tabytac> I like using 1.5x scaling with a 1080p and a1440p monitor
08:25:57 <reldred> sorry, my gpd is 2560x1600, not 1440p
08:26:20 <reldred> it's been my primary machine for the last two and a half months though
08:26:34 <reldred> still don't have a desktop PC over here in brisbong 🙁
08:39:41 *** _zephyris has joined #openttd
08:39:41 <_zephyris> Hmm, should try a pass refining the font for 1.5x scaling... Wonder if it would work. The core problem is that many characters are 5px wide, 2px wide stems and 1px wide counters. 5px * 1.25 = 6.25px, 5px * 1.5 = 7.5px, 5px * 1.75 = 8.75px, ugh.
08:42:46 <peter1138> I'm not sure how hinting works but it feels like a lot of work.
08:43:15 <peter1138> The "gamma" change I have made it look nicer, but the fact that it is not required for any other font suggests that it's the wrong solution.
08:46:55 <peter1138> Having said that, OpenTTD Sans scaled quite nicely in VS Code.
08:48:12 <_zephyris> It should have hinting comprehensive hinting marked, but I don't know anything about how that's actually used to display the font.
08:49:04 <peter1138> However I suspect that VS Code actually renders a larger font and scales it down, because it's almost impossible to get a pixel-perfect font.
08:50:19 <_zephyris> Rendering using hinting has definitely gone out of favour, Word used to, doesn't seem to any more
08:51:14 <peter1138> Hmm, maybe we need to disable hinting.
08:51:35 <peter1138> Actually VS Code at size 20 in VS Code is pretty much perfect.
08:51:57 <peter1138> So the difference is probably down to sub-pixel layouting which OpenTTD definitely does not do.
08:52:18 <peter1138> (We'd need to re-render glyphs for each sub-pixel position.)
08:53:07 <peter1138> I have previously wondered if the string layouting system should in fact handle rendering too.
08:53:42 <peter1138> So instead of rendering "O" "p" "e" "n" "T" "T" "D", we'd render "OpenTTD".
08:54:04 <_zephyris> That'd make the letter shapes inconsistent position to position, right?
08:54:18 <_zephyris> I'd prefer a strong hint over that...
08:54:35 <_zephyris> Always possible that Iv'e got hinting settings wrong anyway
08:54:55 <peter1138> Yes. It is actually desirable for regular fonts, and basically everything else renders that way.
08:55:05 <peter1138> May not be appropriate for OpenTTD Sans etc, though.
08:55:44 <_zephyris> It's a kinda stylistic thing, more sprite-like behaviour, consistent letter shapes.
08:56:05 <peter1138> Like, "inconsistent position" sounds like it would be bad.
08:56:18 <peter1138> But when you see the text properly rendered, you don't notice it because it looks correct.
08:59:31 <peter1138> Of course, it would likely be slower.
08:59:46 <peter1138> But then other software manages to render text when needed without bother.
09:00:35 <peter1138> Though I guess with OpenGL you can move a texture by a sub-pixel for free, and the character by character approach still works.
09:57:13 *** talltyler has joined #openttd
09:57:13 <talltyler> I love having survey data 🙂
10:03:27 <peter1138> Hmm, I have a trigger on DELETE to perform a background update automatically. I just noticed that it gets triggered on any delete, even if no rows are actually deleted. Hmm.
10:19:56 *** dh1 has quit IRC (Ping timeout: 480 seconds)
11:33:45 <_zephyris> Ah, debate on fractional sizing I see...
11:37:20 <_zephyris> FWIW, IMO, the most useful scales are 1.00, 1.25, 1.50, 2.00, 2.50, 3.00, 4.00. That's a roughly log scale, increasing by a factor of 1.26x steps, instead of a linear scale, which _generally_ is how people perceive things like interface scale.
11:50:57 *** firepup has joined #openttd
11:53:39 *** firepup650 has quit IRC (Ping timeout: 480 seconds)
12:22:05 <peter1138> The most useful scale is the one a player picks. I'm not too interested in limiting it to "perceptual steps"
12:22:36 <pickpacket> What even does this function in strings.cpp do? "static void FormatString(StringBuilder &builder, std::string_view str, StringParameters &args, uint case_index = 0, bool game_script = false, bool dry_run = false);"
12:23:00 <ufiby> Non-integer zoom values. I have a resolution of 1920x1080. I use the x1 interface size and Newgrf "U&GUI", and also adjust the fonts manually (font medium_size = 20). OpenTTD itself lacks font scaling, not an interface.
12:24:53 <peter1138> pickpacket, it... formats strings.
12:25:39 <pickpacket> peter1138: but how? That's the entirety of the function definition
12:25:48 <peter1138> pickpacket, that's a declaration. The definition is further down.
12:29:10 <pickpacket> peter1138: oh! Very far down, even
13:20:42 *** Flygon has quit IRC (Remote host closed the connection)
13:57:18 *** SigHunter has quit IRC (Remote host closed the connection)
13:58:18 *** SigHunter has joined #openttd
14:02:58 *** SigHunter has quit IRC (Remote host closed the connection)
14:03:58 *** SigHunter has joined #openttd
14:12:40 <_zephyris> I'm interested in perceptual steps, because it defines where effort is better spent for GUI content. Much easier to optimise a font knowing the game tends to choose 1x, 1.25x, 1.5x, 2x or 3x, and not worrying so much about the 1.75x and 2.5x scales.
14:20:08 *** SigHunter has quit IRC (Remote host closed the connection)
14:21:08 *** SigHunter has joined #openttd
14:37:10 <tabytac> _zephyris: my only worry with that is that ~10% of people do play with e.g 1.75x scaling, and ~19% of people use a scale that isnt in your log defined list.
14:38:28 <tabytac> (-1 uses the suggested scale)
14:38:55 <_zephyris> I'm not saying take the options away, I'm also not saying I'm right!
14:39:27 <_zephyris> My point is when auto-setting scale it's not just about a simple UI scaling, it's also nice to consider the content which needs to be optimised for that scale.
14:40:00 <tabytac> _zephyris: ahh i reread your earlier message that makes sense
14:41:12 *** SigHunter_ has joined #openttd
14:42:11 *** SigHunter has quit IRC (Ping timeout: 480 seconds)
14:57:34 *** locosage has joined #openttd
14:57:34 <locosage> perceptual steps are nice but you need to base them on something adequate and that depends on screen resolution
14:57:49 <locosage> 1x is just too small for pretty much everyone
14:58:10 *** lobster has quit IRC (Ping timeout: 480 seconds)
15:01:36 <locosage> also, need to always keep in mind the context when interpreting survey results
15:01:40 <locosage> as said, there are three kinds of lies: lies, damned lies, and statistics
15:01:58 <locosage> non-integer scales whould've likely be even more popular if they worked better
15:13:16 <_zephyris> And what is the magic -1 hiding 😉
15:13:36 <_zephyris> I do like the new by window size scaling, that's nice
15:15:46 <peter1138> As this is a regular setting, there's no special case for it to see record what is actually used instead of -1.
15:16:30 <peter1138> Could be added I guess.
15:20:23 <_jgr_> The new window size scaling does seem nice, though having the scaling change when you resize the window, e.g. to put two windows side by side, feels a little strange
15:22:14 <_jgr_> That said, using 640x360 as the base size looks way too chunky in my eyes, but I play at 1x so am apparently not in the majority 😛
15:22:15 *** lobstarooo has joined #openttd
15:45:45 <peter1138> Hmm, Minecraft uses only integer scaling, however it's at 2x scaling already at just over 640x480.
15:46:51 *** wallabra2 has joined #openttd
15:48:06 *** wallabra has quit IRC (Read error: Connection reset by peer)
15:50:14 <peter1138> Of course it also has manual control, so you can still set it to 1x at 3840x2160 if you want to.
15:51:39 *** wallabra2 has quit IRC (Remote host closed the connection)
16:02:38 *** wallabra has joined #openttd
16:28:10 *** zanooda2000 has joined #openttd
16:28:10 <zanooda2000> tabytac: yeah, with "custom" fonts it works great - you get not oversized, normal size icons and cursor, but the buttons itself became bigger and easier to aim on. for my 1080p monitor I'm using 1.25x n for steamdeck, as I remember, 1.5x
16:29:10 <zanooda2000> having strict integer scalin with "nice" icon
16:36:03 <ufiby> It's not just the screen resolution, but also the diagonal of the monitor. Samsung's diagonal is 21.5 inches with a resolution of 1920 x 1080 pixels. If the size is equal to or greater than x2, it will be too large. It's a little alarming that there are no font size settings (except for the console or opening the openttd.cfg file via notepad). I do not know why they removed the font size settings
16:39:19 <zanooda2000> probably exactly because `font` command was added
16:43:25 <zanooda2000> so you have harder way but more precise control
16:45:33 <zanooda2000> zanooda2000: or more likely because now font size is related to interface size and then of top of it was added `font` for thoose who don't like auto-scale
16:47:06 <ufiby> There is such a possibility that players can adjust the font size themselves via the console.
16:49:52 <zanooda2000> Yes, but possibility that they don't even know about `font` is way bigger. Especially considering how easy is to adjust your interface size with slider. _Althrough adding a hint about console command would be nice to those who nee but don't know_
17:05:09 *** WormnestAndroid has quit IRC (Ping timeout: 480 seconds)
17:06:05 *** WormnestAndroid has joined #openttd
17:06:31 <ufiby> Yes, it can be. For example, they want to increase the font size, but they don't want to increase the interface size.
18:05:32 <andythenorth> how many railtypes can I have now?
18:06:19 <andythenorth> what if it was 65k, but we blocked out every 12th slot for amusement, as "OpenTTD only"
18:43:29 <peter1138> andythenorth, steps.
19:07:54 <_glx_> Next step might be vehicle type (based on result.txt)
19:21:03 <peter1138> Yeah, there's couple others.
19:23:55 <peter1138> Not sure if there implications for script compatibility though.
19:32:03 *** jessicathegunlady has joined #openttd
19:32:03 <jessicathegunlady> With changing railtypes?
19:33:01 <andythenorth> hmm I should do badges for prop 34
19:36:03 <peter1138> With changing that constant to be -1.
19:36:14 <peter1138> Unit tests that only pass when you modify the unit tests ;)
19:50:15 <peter1138> Hmm, changes to sprite loader make rebasing this not possible.
20:55:05 *** Wolf01 has quit IRC (Quit: Once again the world is quick to bury me.)
21:05:43 *** tokai|noir has joined #openttd
21:05:43 *** ChanServ sets mode: +v tokai|noir
21:12:51 *** tokai has quit IRC (Ping timeout: 480 seconds)
21:13:32 <andythenorth> hmm badges for prop 34?
21:13:36 <andythenorth> or add it to debug
21:13:46 <andythenorth> I did try learning how to extend debug, but I'm a slow learner 😛
21:14:52 <peter1138> Do you mean prop 34 for trains?
21:15:04 <peter1138> Okay. What's the relevance of badges?
21:15:42 <peter1138> What's the relevance of badges?
21:16:17 <andythenorth> if I say 'free debug info' again will you ask the question again? 😛
21:16:55 <andythenorth> there are 2 use cases for wanting to see the vehicle railtype label(s) in game
21:17:04 <andythenorth> 1. did I fuck up my own grf code
21:17:11 <andythenorth> 2. railtype authors probably
21:17:23 <peter1138> Okay, but what's the relevance of badges?
21:18:10 *** mmtunligit has joined #openttd
21:18:10 <mmtunligit> i assume they let you expose that information much more readily than whatever the current way to view it is
21:18:36 <peter1138> Rail type definitions have badges. Vehicle definitions have badges. Copyable badges are copied from rail type definitions to vehicle definitions.
21:18:38 <andythenorth> I am using badges to represent vehicle properties that aren't in the grf debug, in vanilla
21:18:51 <andythenorth> the compile inserts them, using values from the underlying python
21:19:13 <peter1138> What are you actually asking, rather than "badges for prop 34?"
21:19:24 <peter1138> Because that question doesn't make sense. Properties don't have badges.
21:20:04 <andythenorth> ok another better framing
21:21:07 <andythenorth> * I'm quite happy working around limited grf debug using badges
21:21:07 <andythenorth> * is this a bad pattern, should I be filing requests to extend grf debug?
21:21:42 <peter1138> It's a bad pattern, it's not designed for it.
21:24:10 <andythenorth> between this, and `dumpinfo railtypes` a railtype author could reverse lookup the label
21:42:19 *** keikoz has quit IRC (Ping timeout: 480 seconds)
22:44:31 *** dh1 has quit IRC (Quit: My Mac has gone to sleep. ZZZzzz…)
continue to next day ⏵
